The McMath Solar Telescope of
Kitt Peak National Observatory
A. Keith Pierce
The mechanical and optical arrangement of the 160-cm aperture, 90-m focal length, McMath Solar
Telescope is described. Efforts have been made to obtain good images by thermal control of the optical
paths and by selection of fused quartz or metal-based mirrors for the optical components. The optical
paths and performances of a vacuum double-pass spectrometer and single-pass spectrograph are illustrated. Resolution of 600,000 and a total scattered light of 3% are obtained in double-pass.
and height of line formation in
Early in 1954, the National Science Foundation
appointed an ad hoc committee charged with the task
of exploring the desirability and feasibility of a national
observatory operated under the sponsorship of the
This panel, whose members were I. S.
Foundation.
Bowen, L. Goldberg, B. Stromgren, 0. Struve, A. E.
Whitford, and R. R. McMath, chairman, recommended
the immediate construction of two photoelectric stellar
telescopes having apertures of 91 cm and 206 cm (36 in.
and 80 in.), and a large solar telescope with complete thermal control. The Kitt Peak solar telescope was dedicated 2 November 1962 in McMath's honor for the contribution he made to the organization and establish-
and Stark broadening;
the conception of the solar telescope.
servation possible only with a stigmatic spectrograph.
ment of the Kitt Peak National Observatory and in
Introduction
The sun as a source emits radiant energy having a distribution in wavelength which is nearly Planckian in
shape, corresponding to an effective temperature of
5750'K, upon which is superimposed the Fraunhofer
spectrum of nearly 30,000 absorption lines. Each line
is indicative not only of chemical composition but also
of the physical state of the atoms in the solar atmosThe observed
phere forming the absorption line.
positions of the solar lines show wavelength displacements due to: difference in gravitational potential between sun and earth; pressure shifts; mass motions;
and splitting of lines under the influence of magnetic
fields. The observed profiles provide information on:
chemical composition, through the equivalent width
and f values; kinetic temperature and the turbulent
velocity of the atoms along the line of sight; pressure
The author is with the Kitt Peak National Observatory, Tucson, Arizona, which is operated by the Association of Universities
for Research in Astronomy, Inc., under contract with the National Science Foundation.
Received 5 June 1964.
the atmosphere.
For many years solar astronomy dealt with the gross
features of the solar atmosphere: mean wavelengths of
Fraunhofer lines; and models of the photosphere and
chromosphere derived through the assumption of
smooth spherical symmetry and the slowly changing
aspect of sunspots, faculae, plages, and prominences.
Recent work has centered on studies of active solar
regions, flares, and the fine features and inhomogeneities that result from convection, acoustic waves, and
forces. The inhomogeneities
magnetohydrodynamic
are clearly visible if we examine the details of the
Fraunhofer lines along the length of the slit, an obWe then find a complex picture of Doppler shifts and
variations in intensity corresponding to inhomogeneities
in the solar photosphere and chromosphere. -3
A typical photograph of the Doppler shifts made with
the spectrograph of the McMath Solar Telescope at a
time of good seeing is shown in Fig. 1.
The dynamic state of the solar atmosphere has been
4 5
beautifully shown in the work of Leighton et al., ' who
discovered the 5-min oscillation in the phenomena of
6
solar granulation; the observations of Moreton of
blast waves moving out from flares at speeds of 1100
km/sec; and the motion picture evidence for a "wand
waving" motion of filamentary structures in the chromosphere, pushed this way and that by changing radiation
pressure or by dynamical gas forces.
Solar research, therefore, now centers on investigating
the manner in which small-scale inhomogeneities affect
our previously established models of the solar atmosphere. The methods of attack are varied. One means
of deriving information is by high-precision photometric profiles of the Fraunhofer lines and their wave-
lengths obtained point by point along the length of the
slit, area for area on the sun's surface. A solar telescope in conjunction with the spectrohelioscope, or
December 1964/ Vol. 3, No. 12 / APPLIED OPTICS 1337
-
Mt7WNy|t_-
7
7--NyYp
tA-
7-
:T
graph in its Doppler mode, one can show velocity
fields on the sun's surface.' Another useful tool is the
Babcock magnetograph, 0 with which a detailed map of
the general magnetic fields over the sun's surface can be
obtained with a sensitivity of 0.2 G.
In all of these programs, the limit of detection can be
increased or a smaller area studied, or a result obtained
more quickly, making better use of the rare moments of
excellent seeing, provided a large bright image of the
sun is available.
1,
I
I
i
I
I:
A
-E I
i
k
M
I
I
i
I
i
i
i
I
I "1
1i
I
1. I
:
C I
i
j .
f
I
t
Fig. 1. Doppler shifts of Fraunhofer lines produced by rising and
falling currents in the solar atmosphere. The vertical scale of the
figure is 100 see of arc, the horizontal wavelength interval XX5260-
5268.
Fifth order, 1-sec exposure, on Shellburst Linagraphic
70-mm film.
Lyot-Ohman birefringent filter, gives panoramic views
of solar flares and filaments7, to which the addition of
time-lapse cinematography adds a new dimension.
Furthermore, by use of the versatile spectroheliograph
and setting on magnetically sensitive lines, one can
obtain spectroheliograms whose density variations
delineate magnetic fields or, by using the spectrohelio'0
General Design
There were two principal design criteria and several
subsidiary considerations which determined the final
optics and the configuration of the McMath Solar
Telescope. First, it was hoped that the telescope could
be located at a site which would encompass 30 h of 0.5
see of arc seeing, or better, per year. This statement
carries the guiding principal that after the light travels
through the atmosphere the quality of the seeingshould
not be destroyed in the last few hundred feet near the
focus, either by the local terrain or by the telescope
itself. Second, the optical arrangement must be such
as to give 0.33 see of arc resolution. In order to attain
these goals, it was felt that control of the thermal
environment of the telescope was necessary, and that
the possible use of metal mirrors should be investigated.
Many optical arrangements employing mirrors are
possible for a solar telescope: the Cassegrain system
of the usual stellar instrument, coelostat system, heliostat, and siderostat. All of these systems have been
SUN
SECONDARY
McMA2rFE
SQELAR
=rELESCOPDPE
48-INCH MIRROR
AND
MOUNTING
-
F
L,,
ALUMINIZING
HELIOSTAT
SUPPORTTOWER
I UTURE
HORIZONTAL
SPECTROGRAPH
VERTIPCA
L
VACUUM
SFEC'RCORA PH
6,674
T
PIPING
SPECTROGRAP
SHAFT
6,604
Fig. 2.
Cross section of the McMath Solar Telescope.
1338 APPLIED OPTICS / Vol. 3, No. 12 / December 1964
T.
-
e
W2
C KIWI
spectroscopic system with high plate resolution (10 4)
is possible but self-defeating, if one is to do precision
photometry, simply because of plate grain. A more
optimal configuration, from a signal-to-noise ratio
point of view, may be obtained by employing a dispersion which yields a wider projected slit in the image
plane. If we select an f ratio of 60, hence a resolution of 30 ,, a 150 mm X 250 mm grating, then the
spectrograph must have a focal length of 10 m. (We
assume for a solar spectrograph that the collimator and
camera have the same focal length.) We chose to
PLATE
S2
53
overfill the grating and circumscribethe rulings, therefore we selected a 13.7-m focal length.
Having selected an f ratio for the spectrograph collimator, and thus for the telescope, what aperture is
optimum for the telescope? As is well known, a 30-cm
objective will give 0.33 see of arc resolution, and much of
the daytime seeing is far poorer. Several other factors
governed the image size selected and hence the focal
length. First, with an image of the sun approximately
1 m diam, it would be possible at times of very good seeing to do detailed spectrophotometry of granules 1 see of
arc diam (0.5 mm). A program of work on the detailed physical and magnetic fine structure of sunspots
requires a considerable image size. Furthermore, the
use of a Bowen image slicer is indicated for many research projects. An analysis for the sun, similar to
that of Bowen's" for the case of a stellar spectrograph,
0I
0I
0
Ui
CL
with star images greater than the slit width,
shows that the light available to the spectrograph is
M2
MI
M3
Fig. 3. The predisperser and vacuum spectrograph.
The pre-
disperser is a 1-m focal length prismatic spectrograph with a 150
quartz prism. WI and W2 are quartz windows. The 10 cm X 25
cm plate is introduced into the spectrograph through an air lock.
Mirrors M,, M2, and M3 are 30-cm, 30-cm, and 40-cm diam and
13.7-m focal length.
proportional to the first power of the aperture, A t and
the signal to noise is proportional to +/At,. In this case,
the gain goes up slowly, but if one uses an image slicer
the gain in signal to noise goes up directly as the
aperture. Finally, as is to be described later in detail,
we had available two quartz blanks, 160-cm diam,
which, at f160, gave a focal length equal to 100 m.
The Telescope Structure and Mountings
used in the past; each has its advantages and disWe adopted the heliostat because its
advantages.
mounting is simple, involving only a single flat, hence
its cost is less. Also, there is only one reflection, the
polarization and ellipticity of the reflected beam is constant, there is no "noon shadow" of one mirror on the
other as in the coelostat, and, finally, it allowed us to
place the flat high above the ground and away from
thermal disturbances. A heliostat has two disadvantages: it causes the field of view to rotate once in 24
hours, and the angle between incident and reflected
light is in general greater than for the optimum position
of the two mirrors of a coelostat, thus it is more sensitive to mirror figure and astigmatism.
All astronomical instruments are designed around
the dispersing element of the principal spectrograph
and, of course, the research programs proposed for the
spectrograph. Normally, the grating resolving power
and dispersion are matched to the resolution of either
the photographic plate, image tube, or size of the exit
slit. Matching the linear resolving power of the
Figure 2 is an elevation section of the solar telescope
upon which the optical path can be traced. Figures 3
and 4 show the optical path of the vacuum spectrograph. A heliostat, without a second mirror, directs
light alternatively south or north along the polar axis.
A southern direction was preferred since in this arrangement, the heliostat could be placed high above the
ground, as in Fig. 5. Experiments performed at Kitt
Peak to establish the microthermal structure of the
atmosphere showed that the fluctuations decrease
exponentially with height. 12 Near the ground, the range
was found to be 30 C, decreasing to 0.40C at 15-m elevation. We somewhat arbitrarily selected 30-m elevation
for the heliostat, this choice being influenced by factors
of cost and stability of the tower carrying the heliostat.
The design criterion of the heliostat tower specified
that the tower was not to deflect the sun's image more
than 0.33 see of arc when buffeted by a 18 m/sec wind.
This was achieved by surrounding a 9-m diam tower
having 1.2-m thick concrete walls with a windshield,
leaving only the heliostat exposed.
December 1964 / Vol. 3, No. 12 / APPLIED OPTICS
1339
Painting the exterior white using a pure titaniumdioxide pigment in a glyptal vehicle greatly reduced the
refrigeration load. Those panels exposed to the sky
and full sunlight are often much cooler to the touch than
the panels underneath directed toward the ground,
I
I
I
I
-l15C to -30'C.
All mirrors are mounted in carriages which ride on a
3.66-m gauge track extending the full length of the
incline. In this manner, with a suitable hoist, each
mirror can be brought to the aluminizing room for
washing and coating.
No. 1 Mirror Mounting-Heliostat
I
The heliostat was fabricated by Westinghouse
Electric Corporation in their Sunnyvale Plant. This
structure, shown in Fig. 6, consists of an equatorially
mounted yoke into which are set the declination
I
I
trunions that carry the mirror ring and cell. The
I
12,000-kg yoke is supported at the north end by a 44cm spherical, self-aligning roller bearing, mounted at the
I
apex of the north pedestal. The yoke expands at the
I
I
MI
I
I
simply because this paint is white in the visible, black
in the infrared, and the effective sky temperature is
M2
F ig. 4. The double-pass optical system of the Kitt Peak vacuum
spectrograph.
The optical path is inclined to the horizontal 31°
57!5; 60 m of the path are above ground, and 90 m in the
lower optical tunnel are below ground. The return
path directed to the No. 3 mirror is inclined 1.5° below
the polar axis. The portion of the light path above
ground between the heliostat
and ground level is
shielded from the surrounding terrain and sky by a
water-cooled enclosure of square cross section, 10 in of
a side. The incline below ground is cooled by Airtex
panels. A stable nonconvective air column is maintained within the telescope by circulating a coolant in
the panels, from bottom upward, maintaining roughly
a 50 C temperature differential in the 155 m length.
The exterior of the telescope is covered with 30 tons
of copper sheeting. Each removable panel is of dimensions 10 in X 2.5 m; the panel surfaces are subassembled from sheets of copper 40 cm wide containing
three integral, 1-cm diam, tubes which were inflated
south end into a large (304-cm diam) oil-pressure-pad
bearing, through which sunlight reflected from the flat
is projected to the No. 2 mirror. The right-ascension
worm wheel, with 720 teeth, was cut in Westinghouse's
newly adjusted precision gear hobber and then lapped
with the worm in the final assembly.
The right-ascension drive, designed and built by
Boller & Chivens of South Pasadena, consists of a
frequency-controlled
synchronous motor, rated to
drive the heliostat at mean solar time. Steps of 0.05
see of arc may be superimposed on this motion through
a differential and stepping motor. Stepping speeds
varying from 0 to 1500 impulses/sec are available. In
this manner, setting, slow motion, guiding, and various
scan rasters can be programmed into the drive. The
coarse declination drive utilizes a worm wheel. A
2.7 in long tangent arm from the declination axis provides slow motion and guiding, controlled by a stepping
motor in a manner and degree similar to that of the
right ascension system.
The heliostat mirror support presents a problem,
because the mirror normally varies in altitude from
250 above the horizon to 280 below, the latter position
occurring at sunrise in the summer. Instead of the
elaborate counterbalance system often used for telescope
mirrors, we have adopted a pneumatic flotation system.
The mirror itself acts as a piston driven by air pressure
or vacuum proportional to the sine of the altitude of
the normal to the mirror. Additionally, though we
have made provision around the edge for hydrostatic
flotation of the weight of the mirror we have in fact
not yet used it and merely support the edge by a 7-mm
hydrostatically during fabrication of the panel. 70,000
diam rubber 0-ring on the neutral plane of the mirror.
During inclementweather the wholeheliostat mount-
liters of refrigerant water with antifreeze, chilled to a
ing weighing 55,000 kg is lowered 15 m along the incline
suitable temperature, is circulated at 4,000 liters/min
through the structure.
track, and metal doors close off the optical path at the
top of the incline. Restoration of the instrument to
1340 APPLIED OPTICS / Vol. 3, No. 12 / December 1964
Fig. 5. The McMath Solar Telescope. The heliostat is carried at the north end on a 30 m high, 9 m diam, concrete column, surrounded
by a water-cooled windshield. The water-cooled incline carries the track for the mirrors and shields the optical path. The entrance to
the officesand observing room is to the left; the entrance to the aluminizing room is on the right.
operating condition from its stored position requires
approximately 20 min.
No. 2,160-cm Concave Mirror Mounting
The carriage frame is welded from 30-cm heavy-wall
pipe. It supports the 90-m focal-length aluminum
mirror on six pads in a fixed orientation. Focus,
through a 2-m range, is accomplished by moving the
5,000-kg carriage along the track by a motor-driven
ball screw, the nut of which engages a dog in the base
of the focusing knee. The accuracy of this motion depends upon the precise alignment of the rails.
No. 3,122-cm Flat Mirror Mounting
This carriage is smaller than the No. 2 mounting since
it need only support a 122-cm mirror in a fixed location.
Provision has been made for moving the No. 2 and No.
3 carriages to any of three locations along the track.
Light can be directed to three different instruments in
the observing room as desired. The No. 3 mirror size
is determined from the need to see the full solar image,
i.e., to know where one is working on the solar disk with
respect to the limb, and to know the solar coordinates pole, equator, etc.
Solar Telescope Mirrors
Introduction
In most solar telescopes, the experience has been that
within about 2 min after opening the dome and exposing
the mirrors to sunlight the seeing or image quality
One suspects that the factors
markedly deteriorates.3
responsible for the change are either distortion of the
mirrors or convection currents generated over the surface of the mirrors or within the telescope. For a mir-
ror exposedto sunlight, the heating of the surface is not
Fig. 6. The heliostat and auxiliary 1-m heliostats. This photograph, looking north, shows the temporary mirror, 160 cm, in its
cell, counterbalanced by weights at the ends of declination axes.
The oil-pressure-pad bearing and worm wheel are under the 4-m
diam cover at the near (south) end.
inappreciable, for though the mirror may have a high
reflectivity, its emissivity is correspondingly low.
On a clear day at 2500-m elevation, a maximum of
1.6 cal cm- 2 min' is incident on mirrors exposed to
direct sunlight; at the focus of the McMath Solar
Telescope, 6.4 cal cm- 2 min'
is an appropriate figure.
December 1964/ Vol. 3, No. 12 / APPLIED OPTICS 1341
Table I.
Glass
Pyrex
Metals
Fused Quartz
Values of K
K = 152 X 10-5 cgs
50
6
5
An aluminum coating reflects about 90% of the energy,
levaing 10% to be absorbed in the coat and supporting
body. The heat is dissipated by convection and radia-
tion from the front face, conduction into the mirror
body and convection, and radiation and conduction
from the rear face and edge. This asymmetrical heating of the mirror usually destroys its figure.
In a thermally stressed circular plate of thickness
t, one face of which is at temperature T and the
other T + AT, and for a uniform temperature gradient
the plate will assume a spherical curvature equal to
t/a X AT, where a is the coefficient of expansion of the
material. 1 For example, for a quartz mirror 150-cm
diam X 20 cm thick to remain flat to IX the temperature
differential front to back must be less than 0.06'C.
Couder 5 has compared different mirror materials.
He remarks that temperature differences introduce
4
distortions proportional to K = acc/m, where a is the
coefficient of expansion,
the density, c the specific
heat, and m the heat conductivity. Table I lists values
of K. Because of the inverse relationship between a
and m, it is apparent that in spite of the great difference in coefficients of expansion between fused
quartz and metal, the two are on a par with each other
and either might be acceptable as a telescope mirror
material. Of course, the advantage of the metal
mirrors in place of quartz mirrors in a solar telescope
resides in the possibility of more easily cooling the
metal mirror by internal ducts, and its shorter time
constant with respect to temperature changes.
Quartz Mirrors
In 1953,through the interest of Ira S. Bowen,director
of the Mount Wilson and Palomar Observatories, the
quartz mirrors produced in 1932 for the "200-inch"
project by the General Electric Company were made
available to McMath for a future large solar telescope.
Some years later they were presented to AURA for use
in the Kitt Peak telescope, and much of our conceptual
planning revolved around these blanks. There were
three disks: (a) 160-cm diam X 24 cm thick, this
thickness being made up of a 5-cm thick layer of
relatively clear quartz on an opaque bubble-filled back;
(b) A second disk similar to (a) 165-cm diam X 21-cm
thick, which had cracked diagonally in halves and had
been welded back together.
Its back surface,
particularly near the edges, was in poor shape, as
rather large chunks could be removed easily by hand.
This disk was later sawed to 122-cm diam X 18-cm
thick. (c) A third disk with no quartz overcoating,
consequently unsuitable for mirrors.
Mirror (a) became the heliostat flat, and (b) the
No. 3 flat of the solar telescope. These mirrors, insofar
as has been observed, have maintained their figures at
1342 APPLIED OPTICS / Vol. 3, No. 12 / December 1964
all times, i.e., we have found no astigmatism attributable to these mirrors; and they would appear to be
excellent for a solar telescope, except that there are
many surface defects which scatter light and degrade
the contrast. Whether convection from the faces of
these mirrors has contributed to the occasional poor
seeing has not yet been determined. These mirrors
will soon be replaced with other blanks of either quartz
or metal.
Metal Mirrors
It has been our hope that the use of metal mirrors
with their high thermal conductivity, together with the
possibility of cooling them to ambient temperature,
would eliminate poor images produced by convection
currents of air rising from the mirror face, if indeed
this is a factor. Hence, we have undertaken a program of development of metal mirrors, recognizing
that the principal problem probably would be the
question of the long-term stability of the mirror blank.
We have tried mirrors, 26-cm diam, of Fe, Cu, Be
and Al. Each mirror was overcoated with a 130-A
layer of Kanigen (trade name for an amorphous
nickel-phosphorous alloy) . 1 Kanigen is a hard material,
which can be ground and polished by the usual optical
techniques. Our experience has been that Fe, Be, and
Al blanks appear to be suitable for mirrors, but that
Cu is too soft a base material. Most of our work has
been with 356-T6 aluminum casting alloy, and mirror
blanks of this material of 40-160 cm have been
produced. The No. 2 mirror of the solar telescope is
an aluminum casting, 160-cm diam and 25.4 cm over-all
thickness, ribbed in a triangular-hexagonal pattern,
each triangle being 18 cm high with 2.5 cm thick ribs,
see Fig. 7. This casting, weighing 710 kg, has a face
thickness of 3.8 cm and is overcoated with 130 4 of
Kanigen. It was machined to nearly the proper concave shape, then rough ground to a sphere, Kanigen
coated, fine ground, and polished to a concave sphere of
Fig. 7. Back of the aluminum casting, 160-cm diam, 25.4 cm
thick. This photograph shows the mirror in its test-rig, held by
a stainless-steel band. The small, sponge-rubber pads cemented
to each triangular intersection are for support of the mirror on the
polishing machine.
Finally, after figuring, it was
88-m focal length.
aluminized to increase the reflectivity.
In spite of an excellent polish and freedom of scattering, one of the problems of figuring Kanigen-coated
metal mirrors has been a surface ripple or irregularity
having a spacing of 5-20 cm, and only a fraction of
a wavelength amplitude. Our polishing techniques
have not yet reached the stage where we can obtain as
regular a surface as with fused silica.
Telescope Performance
It has proved difficult to decide between fused
quartz or metal mirrors for the final optics of the solar
telescope. There appears to be no reasonable doubt
as to the long-term stability of quartz. We cannot yet
say the same for metals, particularly alloys. In order
to investigate further the relative merits of each, it is
our intention to install a fused quartz mirror along with
the present aluminum mirror. Both would be mounted
on a turntable so that they could be rapidly interchanged. Thus, we can directly compare their behavior, one with respect to the other, for stability and
for the effects of cooling. The best images obtained
from the aluminum mirror, in combination with the
heliostat and No. 3 mirror, as evaluated from stellar
observations, focuses 100% of the light within a 2 see of
arc circle and 50% within 1 see of arc. Seeing and
mirror figure combine to produce a smearing of the
image. From the profile of the sun's limb, on a day of
moderately good seeing, W. C. Livingston has measured
this image smear photoelectrically. He finds it is
nearly gaussian in shape with a width, or standard
deviation,
of 1.5 see of arc.
The internal seeing in the telescope is examined with a
small telescope, which is focused on the heliostat
through the optical train. Often there are quiet days,
when the dust particles are nearly motionless in the
light beam; these days have excellent internal seeing
with tremor less than 0.5 see of arc. It is observed
that large diurnal temperature changes with some wind
can give poor internal seeing, but again, more observational data is needed.
On several occasions, at night, star images have been
stable to 0.2 see of arc, though the image size, as noted
before, is much larger than this. For daytime seeing
on rare occasions when conditions are best, one to two
hours after sunrise, the polygonal nature of the granulation is clearly visible but more often only the speckled
appearance of the 1 see granules is visible. At times
of best seeing the focus is determinable to about i 2
cm.
Vacuum Spectrograph
Modern solar spectroscopy encompasses as one of its
goals the precision determination of line profiles.
The equivalent width of a Fraunhofer line, the halfwidth, the asymmetry, and the central intensity are all
parameters which can be related to the physical processes of line formation and to the problem of radiative
transfer in the solar atmosphere. For example, asymmetries can be due to pressure effects or to Doppler
shifts, arising from convection in the outer solar layers
in which bright, hot elements of the gas rise, and cool,
darker material descends. Some of the effects under
study have a very small influence on the line profiles;
hence, the necessity of making intensity measurements
with high precision, say 0.01% in intensity.
To approach this goal, the spectrograph must possess the highest attainable resolution, and great pains
must be taken in determining or eliminating the effect
of the instrumental profile, ghosts, and scattered light.
The Optical System
If only short wavelength intervals are to be considered or the instrument is to be used as a spectrometer,
an all-reflecting system is to be preferred, for it avoids
the need of refocusing for different wavelengths, and
diaphragms can be placed to catch much of the
internally scattered light. We have copied essentially
the modified Czerny optical system used in the Mc-
Math-Hulbert vacuum spectrograph.17 The light paths
are illustrated by the schematic shown in Fig. 3.
Nearly all the photographic work is done in single pass.
Though the photographic system is noncompensating
for astigmatism, as in the Czerny-Turner arrangement,
because of the large focal ratio, f/60, the image quality
is excellent. Photographs of the iodine band structure
near 5330 show clearly resolved lines of measured
separation 0.010 A for fifth-order spectra having a
linear dispersion of 7.5 mm/A.
Photoelectric scanning can be carried out either in the
single- or double-pass modes. It is worthwhile to
describe our first attempt at double passing the vacuum
spectrograph, although it was a failure from our point
of view. The optical path used is illustrated in Fig. 8.
Light from one half the length of the entrance slit,
S2, followed through in single pass to the intermediate
defining slit, S. It was returned through the system
by mirror, M4, and emerged to the side of the entrance
slit body at S3, filling the field of view corresponding to
the other half of the slit length. However, in this arrangement, light from nearby wavelengths also fell in
the Littrow mode on the collimator, which in turn
focused the spectrum on the region of the entrance slit,
S2, and, in particular, also on the grating. From this
point, the light diverged from the grating again filling
the collimator, M1 , and from there was sent as a parallel
bundle of light to the region of the entrance slit, S2, M5,
flooding the field of view. Had we chosen to chop the
beam at the intermediate slit and recorded only the
modulated output, the scheme would have succeeded.
However, we felt it far preferable to adopt another arrangement which has worked to our entire satisfaction.
By tilting mirror, M2, shown in Fig. 4, light is sent to
a pair of mirrors, M3 and M4, off the plane of the system,
and returned a second time through the optics to the
exit slit. In this double-pass arrangement there
are no virtual images, and it is possible to baffle the
system completely. If long wavelength intervals are
desired, the spectrum is scanned by rotating the grating;
if short scans are desired, the spectrum is scanned by
December 1964 / Vol. 3, No. 12 / APPLIED OPTICS 1343
Most gratings show a slight amount of astigmatism,
and a run or change of focus with order. Our particular
grating is no exception, as shown by Fig. 11. Il-
luminating the grating with parallel light, collimator at
zero setting, the dust streaks and spectral lines are
separated by several centimeters, and the separation is
II
\\I
I5
variable with order. Lord Rayleigh and others-1 2 0
have given an analysis of these errors for plane gratings.
The astigmatic foci can be brought into approximate
coincidence by changing the position of the collimator,
by 12.5 cm, so nonparallel light is directed to the
I
The Mechanical System
grating.
\\I
\'
It is desirable to be able to orient the slit of the
\ l I
\I
\
spectrograph in different position angles, either with
respect to the limb or with respect to sunspots or plages
on the sun's surface. Though this could be accomplished through an image rotator, we felt, because of the
large scale of the primary image, the additional reflections, the problems of maintaining adjustment of the
rotator, and because of the rotation of the field of view
once in 24 h, that it would be better to attempt a moving
spectrograph. This seemed particularly feasible as the
I
I
I
spectrograph could be set vertically, thus the gravitaM2
Mj
Fig. 8. First attempt at double pass, which failed because of
scattered light from the grating.
moving the exit slit and the intermediate slit in synchronism along the line of dispersion.
tional deflections would be constant.
The spectrograph optics are mounted in a 2-m diam,
21 m long, steel tank, weighing 17 metric tons. The
bottom end carries a spherical cap 56-cm diam resting
in a cup. A dynamic oil film, maintained by a flow of
4 liters/min, holds the separation of the bearing faces
l
The Grating
60
Our 610 grooves/mm grating, by Horace W. Babcock, is ruled on an aluminum-coated blank 31-cm diam
X 5 cm thick. The grating has a ruled area 25-cm X 15
cm (N = 155,000). The remarkable blaze of this
50
40
30
20
10
grating in the double-pass arrangement, the square of
the single pass, is illustrated by Fig. 9.
200
The figure rep-
90
80
resents the intensity in the solar spectrum observed
with a photomultiplier.
The factors of energy distribution and spectral response of the receiver are not removed from the graph.
We have used several tests for resolving power.
Direct photography shows a resolution in the fifth
order of approximately 500,000 for the components
X5641 of Hg and for the lines of I2, Fig. 10, in the singlepass mode. In the double-pass arrangement, the corresponding figure is 600,000. These figures are not
corrected for slit width or line width. Similar values
are obtained by photoelectric scanning of the spectrum.
Visually, the resolution approaches very closely the
Rayleigh values 775,000 and 1,550,000, respectively.
The integrated intensity of all the ghost structures
amounts to about 5% of the central peak. Additionally, there is 8% general scattered light, hence, the
need for the double-pass system when measuring absorption line intensities. In double pass, the total
scattered light is 3% when referred to the absolute zero
condition of "shutter closed".
1344 APPLIED OPTICS / Vol. 3, No. 12 / December 1964
70
60
30
40
30
20
10
I-
IL
100
90
Z
80
70
60
50
40
30
20
10
TH
9
3000
4000
5000
WAVE
6000
7000
LENGTH
Fig. 9. The blaze of the grating as observed in double pass in
different orders. These curves were made as constant slit width,
and include the response of the EMI 9558 cell and the energy distribution of the sun, which peaks at X5000.
provide for 1 in change of focus. Because of their inaccessibility, electrical controls at the head of the spectro-
graph are provided to permit adjustment of each mirror
for tilt, collimation, and focus.
Many grating drives have been devised, some use
sine arms, others tapes; we have adopted the latter.
B
b c
a
Fig. 10. 12 absorption lines superimposed on the Fraunhofer
spectrum. A is the fifth-order spectrum in single pass; B is the
fifth order in double pass. Components a of mean wavelength
X5328.904 have a measured separation of 0.0090 A, b = 0.0104 A,
and c = 0.0100 A. Film: contrast process ortho. Slit: 30 ,c,
10 see exposure.
cell at 310'K.
5 cm long iodine absorption
Figure 12 explains the mechanism of rotation of the
grating. The grating is supported in a cell, which in
turn is defined by a system of push-pull screws to allow
the grooves of the grating to be adjusted exactly
parallel to the axis of rotation. The cell is carried in a
large aluminum spool, whose wheels are 76.2-cm diam.
An internal slewing gear couples the grating spool to
auxiliary wheels at each end, to which tapes are attached. The thin ribbon tapes leave each wheel
tangentially, and are fastened to an equalizer arm which
carries at its center the nut of the precision drive screw.
Two other tapes support 1-kg counterweights, which
maintain constant pressure on the drive tapes and preload the screw. The 2.5-cm diam screw has twenty
threads per cm. It was carefully lapped, and a sapphire thrust bearing inserted in the end to bear against
a 1-mm diam graphitar thrust point.
The grating
spool is supported at each end by precision ball bearings, held in a welded framework that extends down from
the lid of the spectrograph tank. An extension of the
grating rotation axis carries a finely divided circle,
allowing one to set visually the grating angle to 1 min
of arc.
20
r
15
0lL
C,
0
X
COLLIMATORA
0
11
U-
0
w
10
N
5
0
COLLIMATOR AT
I
190
0
II
II
210
230
250
270
290
GRATING ANGLE
Fig. 11. Astigmatic behavior of the grating in different orders.
at 75 ,u to 125 ,u, and floats the weight of the tank and
optics. Side supports near the top of the tank are
simple rollers riding against a ring. This same ring
is wrapped by a 2.5-cm wide tape that also passes over
two drive wheels, 0.10 diam of the tank ring, thus
providing a friction drive for the tank rotation. The
bottom bearing carries a tubular extension, and an
0-ring seal through which the tank is evacuated. The
mirror mounts for the camera and collimator mirrors
Fig. 12. The grating support and drive hang from the heavily
braced top plate of the spectrograph. For slew, the Bodine
motor drives the inner spool with respect to the tape-driven
wheels. Slow scan is from the tapes attached to the equalizer
arm and screw driven by a shaft through the top of the tank.
December 1964 / Vol. 3, No. 12 / APPLIED OPTICS 1345